Boundedly Rational Agents Research Articles

Organizational Coordination and Costly Communication with Boundedly Rational Agents

How does costly communication affect organizational coordination? This paper develops a model of costly communication based on the weakest-link game and boundedly rational agents. Solving for the stochastically stable states, we find that communication increases the possibilities for efficient coordination compared to a setting where agents cannot communicate. But as agents face a trade-off between lowering the strategic uncertainty for the group and the costs of communication, the least efficient state is still the unique stochastically stable one for many parameter values. Simulations show that this is not just a long run phenomena, the stochastically stable state is the most frequent outcome also in the short run. Making communication mandatory induces efficient coordination, whereas letting a team leader handle communication increases efficiency when the leader expects others to follow and has enough credibility. The results are broadly consistent with recent experimental evidence of communication in weakest-link games.

Preferential Attachment and the Search for Successful Theories

Multiarm bandit problems have been used to model the selection of competing scientific theories by boundedly rational agents. In this article, I define a variable-arm bandit problem, which allows the set of scientific theories to vary over time. I show that Roth-Erev reinforcement learning, which solves multiarm bandit problems in the limit, cannot solve this problem in a reasonable time. However, social learning via preferential attachment combined with individual reinforcement learning, which discounts the past, does.

Irrational fads, short-term memory emulation, and asset predictability

Opponents of the efficient markets hypothesis argue that predictability reflects the psychological factors and “fads” of irrational investors in a speculative market. In that, conventional time series analysis often fails to give an accurate forecast for financial processes due to inherent noise patterns, fat tails, and nonlinear components. A recent stream of literature on behavioral finance has revealed that boundedly rational agents using simple rules of thumb for their decisions under uncertainty provides a more realistic description of human behavior than perfect rationality with optimal decision rules. Consequently, the application of technical analysis in trading could produce high returns. Machine learning techniques have been employed in economic systems in modeling nonlinearities and simulating human behavior. In this study, we expand the literature that evaluates return sign forecasting ability by introducing a recurrent neural network approach that combines heuristic learning and short-term memory emulation, thus mimicking the decision-making process of boundedly rational agents. We investigate the relative direction-of-change predictability of the neural network structure implied by the Lee–White–Granger test as well as compare it to other well-established models for the DJIA index. Moreover, we examine the relationship between stock return volatility and returns. Overall, the proposed model presents high profitability, in particular during “bear” market periods.

Evolution and market behavior in economics and finance: introduction to the special issue

The articles contributing to this special issue were originally presented at a Workshop held in September 2009 at the Sant’Anna School of Advanced Studies in Pisa and titled “Evolution and Market Behavior in Economics and Finance”. The aim of the workshop was to gather scholars working on the application of evolutionary theory concepts, such as selection, mutation and replication, to economics, in particular, to the understanding of the selection operated by markets, and to discuss ideas for future research in this area. The analysis of market selection and the characterization of the emerging market properties, such as allocative and informational efficiency, from an “evolutionary” perspective, have received an increasing attention in the last years. The typical evolutionary model studies the exchange of real or financial assets by heterogeneous boundedly rational agents, using the evolution of cumulated wealth or payoff to determine survival and dominance of specific strategies or behaviors and, consequently, the aggregate economic dynamics. The central question is indeed which types of expectations, information sets, preferences, learning models are rewarded in the long-run and their effect on market properties. More generally, by considering an a-priori large ecology of behaviors, evolutionary market models are good candidates to question the dominating neoclassical paradigm embodied by the use of households and firms characterized by “substantive rationality” (Simon 1976), that is, optimizing with respect to a well defined

Role models that make you unhappy: light paternalism, social learning, and welfare

AbstractBehavioral (e.g., consumption) patterns of boundedly rational agents can lead these agents into learning dynamics that appear to be ‘wasteful’ in terms of well-being or welfare. Within settings displaying preference endogeneity, it is however still unclear how to conceptualize well-being. This paper contributes to the discussion by suggesting a formal model of preference learning that can inform the construction of non-standard notions of dynamic well-being. Based on the assumption that interacting agents are subject to two biases that make them systematically prefer some cultural variants over others, we develop a procedural notion of well-being, based on the idea that policy should modify institutional conditions that generate dynamic instability in preference trajectories, while leaving individual choice sets unrestricted.

Monetary Policy, Banking and Heterogeneous Agents

The influence of heterogeneous expectations on monetary policy performance has gained a lot of attention in the recent years. It proved to be an important factor that, under some circumstances, may even destabilize the economy (Massaro, 2012). This paper investigates the phenomenon of heterogeneous expectations further, analyzing its role in the monetary policy conduct with an active banking sector. In our analysis we assume a constant fraction of boundedly rational agents who use simple heuristics to form their expectations. The impact of those biased beliefs is studied in the framework, originally developed by Goodfriend and McCallum (2007). We first show that the presence of the banking sector changes the determinacy structure of the system and, depending on the heuristics used, the presence of the boundedly rational agents might have either stabilizing or destabilizing effect. In particular, when boundedly rational agents extrapolate the past performance in forming their expectations, the range of the stable (determinate) monetary policy instruments is narrowed.

Social Learning with Coarse Inference

We study social learning by boundedly rational agents. Agents take a decision in sequence, after observing their predecessors and a private signal. They are unable to make perfect inferences from their predecessors' decisions: they only understand the relation between the aggregate distribution of actions and the state of nature, and make their inferences accordingly. We show that, in a discrete action space, even if agents receive signals of unbounded precision, there are asymptotic inefficiencies. In a continuous action space, compared to the rational case, agents overweight early signals. Despite this behavioral bias, eventually agents learn the realized state of the world and choose the correct action. (JEL D82, D83)

Imitating cooperation and the formation of long-term relationships

We propose that the option to maintain or to quit relationships can enforce cooperation in the prisoner’s dilemma between boundedly-rational agents in a large population if players imitate successful behavior and sometimes choose to experiment. Agents do not have the ability to follow history-dependent strategies, but have access to information about what behavior is on average the most successful one. Cooperative behavior in a homogenous population emerges if both the imitation and experimentation rate are suciently small. In a heterogeneous population, the same holds without a restriction on the imitation rate. The results resemble Eshel et al. (1998), but the assumption of local interaction is substituted by the endogenous formation of relationships.

A Model of Persuasion with Boundedly Rational Agents

A new model of mechanism design with a boundedly rational agent is studied. A speaker presents a request to a listener who would like to accept the request only if certain conditions are met by the speaker’s true profile. This persuasion situation is modeled as a leader-follower relationship. The listener first announces and commits to a persuasion rule, i.e., a set of conditions to be satisfied by the profile in order for him to be persuaded. Then, the speaker presents a profile, though not necessarily the true one. The speaker is boundedly rational in the sense that his ability to come up with a persuasive profile is limited and depends on the true profile and on the persuasion rule and the way in which it is framed. We fully characterize the circumstances under which the listener’s goal can be achieved.

Major trends in agent-based economics

The study of the economy by means of Agent-Based (AB) models is a relatively new field. It dates back to the early 90’s, when the increasing availability of cheap computing power has made possible to undertake the first computationally demanding experiments required tomodel the interactions of a large number of boundedly rational and heterogeneous agents (for a review, see Tesfatsion and Judd 2006), in an economy characterized by non-equilibrium dynamics and information asymmetries. The AB approach allows us to build models with a large number of heterogeneous agents, where the resulting aggregate dynamics is not known a priori, and outcomes are not immediately deducible from individual behaviour. This approach is characterized by three main tenets: (i) there is a multitude of objects that interact with each other and with the environment; (ii) objects are autonomous (hence, they are called agents), no central or “top down” control over their behaviour is admitted; and (iii) the outcome of their interaction is numerically computed. During the last decade, a growing number of research papers using agent-based models has appeared in the literature, focusing on different fields of economic theory, such as finance (e.g., Raberto et al. 2001; LeBaron 2006), industrial organization (Kutschinski et al. 2003), labour market (Tassier and Menczer 2001), innovation (see e.g. Dawid 2006), and the relationship between financial fragility of firms and business

Freedom of choice and bounded rationality: a brief appraisal of behavioral economists' plea for light paternalism

Behavioral economics has addressed interesting positive and normative questions underlying the standard rational choice theory. More recently, it suggests that, in a real world of boundedly rational agents, economists could help people to improve the quality of their choices without any harm to autonomy and freedom of choice. This paper aims to scrutinize available arguments for and against current proposals of light paternalistic interventions mainly in the domain of intertemporal choice. It argues that incorporating the notion of bounded rationality in economic analysis and empirical findings of cognitive biases and self-control problems cannot make an indisputable case for paternalism.

CATEGORIZE THEN CHOOSE: BOUNDEDLY RATIONAL CHOICE AND WELFARE

We propose a boundedly rational model of choice where agents categorize alternatives before choosing. The model explains some behavioral anomalies, and it is fully characterized by a property of choice data: a categorizer can never exhibit certain patterns of “revealed preference reversals”. This model offers clues on the problem of making welfare judgements in the presence of boundedly rational agents.

Heterogeneous expectations and strong uncertainty in a Minskyian model of financial fluctuations

We examine the role of expectations in a model aimed to explain financial fluctuations. The model restates the core of Minsky's financial instability hypothesis, focusing on the role of expectations. The hypotheses concerning the process of formation and revision of expectations are discussed in light of Keynes's epistemological view of the behavior of boundedly rational agents under conditions of strong uncertainty. These hypotheses are formalized by drawing on recent advances in complex dynamics, decision theory and behavioral economics. We show that widespread use of extrapolative expectations by economic agents produces a high degree of financial instability that may lead to a serious financial crisis, and that the use by economic agents of a mix of extrapolative and regressive expectations reduces the dynamical instability of the model but may give rise to complex dynamics.

Behavioural Finance and Financial Economics with Heterogeneous Beliefs

The literature about the dynamics of prices in speculative markets, based on the interaction of boundedly rational heterogeneous agents has become well developed in recent decades. Hommes CH [1] provides an excellent survey work on Heterogeneous Agent Models (HAMs) in economics and finance. In this strand of literature, one is able to generate sophisticated structures that capture some of the dynamics and stylized facts documented in financial time series at the macro level by aggregating the simple interactions of boundedly rational agents using various trading rules at the micro level. Such stylized facts include excess volatility, high trading volume, temporary bubbles and trend following, sudden crashes and mean reversion, clustered volatility and fat tails in the returns distribution. These models generally include nonlinear elements which may come from agents’ or demand functions, evolutionary switching between available strategies, and contagion and consequent transition of speculators among optimistic and pessimistic groups.

Agent-based dynamics in the general equilibrium model

We introduce agent-based dynamics in a class of Arrow-Debreu economies with capital accumulation and technological progress. In this framework we confirm the results obtained by Gintis (2006, 2007) in an exchange economy: the micro-behavior of boundedly rational agents can lead to the emergence of equilibrium at the macro-level provided some form of collective optimization takes place via evolutionary mechanisms. Sensitivity analysis hint at further properties of out-of-equilibrium dynamics which can be analyzed using agent-based models.

Does cumulative advantage affect collective learning in science? An agent-based simulation

Agent-based simulation can model simple micro-level mechanisms capable of generating macro-level patterns, such as frequency distributions and network structures found in bibliometric data. Agent-based simulations of organisational learning have provided analogies for collective problem solving by boundedly rational agents employing heuristics. This paper brings these two areas together in one model of knowledge seeking through scientific publication. It describes a computer simulation in which academic papers are generated with authors, references, contents, and an extrinsic value, and must pass through peer review to become published. We demonstrate that the model can fit bibliometric data for a token journal, Research Policy. Different practices for generating authors and references produce different distributions of papers per author and citations per paper, including the scale-free distributions typical of cumulative advantage processes. We also demonstrate the model’s ability to simulate collective learning or problem solving, for which we use Kauffman’s NK fitness landscape. The model provides evidence that those practices leading to cumulative advantage in citations, that is, papers with many citations becoming even more cited, do not improve scientists’ ability to find good solutions to scientific problems, compared to those practices that ignore past citations. By contrast, what does make a difference is referring only to publications that have successfully passed peer review. Citation practice is one of many issues that a simulation model of science can address when the data-rich literature on scientometrics is connected to the analogy-rich literature on organisations and heuristic search.

A Class of Evolutionary Models for Participation Games with Negative Feedback

We introduce a framework to analyze the interaction of boundedly rational heterogeneous agents repeatedly playing a participation game with negative feedback. We assume that agents use different behavioral rules prescribing how to play the game conditionally on the outcome of previous rounds. We update the fraction of the population using each rule by means of a general class of evolutionary dynamics based on imitation, which contains both replicator and logit dynamics. Our model is analyzed by a combination of formal analysis and numerical simulations and is able to replicate results from the experimental and computational literature on these types of games. In particular, irrespective of the specific evolutionary dynamics and of the exact behavioral rules used, the dynamics of the aggregate participation rate is consistent with the symmetric mixed strategy Nash equilibrium, whereas individual behavior clearly departs from it. Moreover, as the number of players or speed of adjustment increase the evolutionary dynamics typically becomes unstable and leads to endogenous fluctuations around the steady state. These fluctuations are robust with respect to behavioral rules that try to exploit them.

Endogenous Equilibria in Liquid Markets with Frictions and Boundedly Rational Agents

In this paper we propose a simple binary mean field game, where N agents may decide whether to trade or not a share of a risky asset in a liquid market. The asset's returns are endogenously determined taking into account demand and transaction costs. Agents' utility depends on the aggregate demand, which is determined by all agents' observed and forecasted actions. Agents are boundedly rational in the sense that they can go wrong choosing their optimal strategy. The explicit dependence on past actions generates endogenous dynamics of the system. We, firstly, study under a rather general setting (risk attitudes, pricing rules and noises) the aggregate demand for the asset, the emerging returns and the structure of the equilibria of the asymptotic game. It is shown that multiple Nash equilibria may arise. Stability conditions are characterized, in particular boom and crash cycles are detected. Then we precisely analyze properties of equilibria under significant examples, performing comparative statics exercises and showing the stabilizing property of exogenous transaction costs.

Pareto-Optimal Matching Allocation Mechanisms for Boundedly Rational Agents

This article is concerned with the welfare properties of trade when the behavior of agents cannot be rationalized by preferences. I investigate this question in an environment of matching allocation problems. There are two reasons for doing so: firstly, the finiteness of such problems entails that the domain of the agents' choice behavior does not need to be restricted in any which way to obtain results on the welfare properties of trade. Secondly, some matching allocation mechanisms have been designed for non-market environments in which we would typically expect boundedly rational behavior. I find qualified support for the statements that all outcomes of trade are Pareto-optimal and all Pareto optima are reachable through trade. Contrary to the standard case, different trading mechanisms lead to different outcome sets when the agents' behavior is not rationalizable. These results remain valid when restricting attention to minimally irrational behavior.

Exchange rate determination, macroeconomic dynamics and stability under heterogeneous behavioral FX expectations

Abstract In this paper the role of behavioral forecasting rules of chartist and fundamentalist type for the dynamic macroeconomic stability of a two-country system is investigated. The main result of the paper is that for large trend-chasing parameters in the chartist rule used in the FX market, not only this market but also the entire macroeconomic system is destabilized. This outcome takes place despite of a monetary policy conduction in both countries which satisfies the Taylor Principle, and thus highlights the limits of monetary policy as a macroeconomic stabilization instrument in a world of boundedly rational agents.